| Περίληψη: | Due to technological advances in hardware and an attempt to improve flexibility in production cells, robots no longer require to be isolated behind safety fences, but rather interact with operators. Human – Robot Collaboration (HRC) is a widely adopted concept in the industrial realm. Physical interaction between operators and robots has enabled, the former to focus on high value tasks and the latter to complete repetitive, physical strenuous and risky tasks. This thesis proposes a robust framework to co–manipulate the folding of fabrics. The framework utilizes the Kinect RGB/D sensor to identify the position and orientation of the fabric using image processing to extract geometric features. The same sensor is used to track the position and orientation of the operator’s hand inside the collaborative space. To improve the Kinect’s tracking algorithm and eliminate noise during the collaboration, a non – linear Kalman filter is applied to the Kinect’s tracking data. In addition, the constructed framework consists of the KUKA LWR IV+ 7-axis collaborative robot, equipped with an ATI force/torque sensor and a Reflex One 3 finger – robotic gripper with 5 DOF that resembles the human hand movements. The robot reacts to the operator’s movement accordingly and is guided solely by the human operator. The proposed system was tested in a real use case scenario, where the collaborative task is the fold of a rectangular fabric. The fabric is laid flat on a table inside the collaborative space and the robot task is to follow and assist the human operator to fold the fabric in two different directions, along the two different fabric dimensions. The experimental results showed that the proposed system is able of identifying the laid fabric, tracking the human position and orientation and guide the robot to assist in the fabric folding.
|
|---|
